Guilt, shame, and postpartum infant feeding outcomes: A systematic review

Abstract Negative maternal affect (e.g., depression and anxiety) has been associated with shorter breastfeeding duration and poorer breastfeeding intention, initiation, and exclusivity. Other affective states, including guilt and shame, have been linked with formula feeding practice, though existing literature has yet to be synthesised. A narrative synthesis of quantitative data and a framework synthesis of qualitative and quantitative data were conducted to explore guilt and/or shame in relation to infant feeding outcomes. Searches were conducted on the DISCOVER database between December 2017 and March 2018. The search strategy was rerun in February 2020, together yielding 467 studies. The study selection process identified 20 articles, published between 1997 and 2017. Quantitative results demonstrated formula feeders experienced guilt more commonly than breastfeeding mothers. Formula feeders experienced external guilt most commonly associated with healthcare professionals, whereas breastfeeding mothers experienced guilt most commonly associated with peers and family. No quantitative literature examined shame in relation to infant feeding outcomes, warranting future research. The framework synthesis generated four distinct themes which explored guilt and/or shame in relation to infant feeding outcomes: ‘underprepared and ineffectively supported’, ‘morality and perceived judgement’ (breastfeeding), ‘frustration with infant feeding care’ and ‘failures, fears and forbidden practice’ (formula feeding). Both guilt and shame were associated with self‐perception as a bad mother and poorer maternal mental health. Guilt and shame experiences were qualitatively different in terms of sources and outcomes, dependent on infant feeding method. Suggestions for tailored care to minimise guilt and shame, while supporting breastfeeding, are provided.


| INTRODUCTION
Breastfeeding provides health benefits to infants, such as reduced risk of infectious morbidity and mortality, dental malocclusions, and overweight and diabetes later in life (Victora et al., 2016). Breastfeeding also protects mothers from breast and ovarian cancer and reduces risk of diabetes (Victora et al., 2016). As such, the World Health Organisation (WHO) recommend exclusive breastfeeding for the first 6 months postpartum (UNICEF, 2017). Despite awareness of breastfeeding benefits and promotional campaigns (Thomas, 2014), WHO recommendation compliance remains poor in developed countries.
A systematic review involving 11 European countries completing a standardised national survey found that in all participating countries, breastfeeding rates declined gradually from initiation after birth to 1 to 2 months postpartum and at 6 months postpartum (e.g., in the Netherlands, from 80% initiation, to 64% prevalence at 2 months postpartum, and 51% prevalence at 6 months postpartum ;Theurich et al., 2019). A similar decline can be seen between breastfeeding initiation and breastfeeding duration in other developed countries, including Australia (Australian Government: Department of Health, AGDH, 2019), Canada (Chalmers et al., 2009), United Kingdom (McAndrew et al., 2012 and the United States (Center for Disease Control and Prevention, CDCP, 2019). Given these trends, it is important to explore potential factors contributing to the gap between breastfeeding initiation and breastfeeding prevalence at 6 months postpartum in developed countries.
Maternal emotional state is a modifiable factor which affects breastfeeding outcomes. In a systematic review of 48 studies, higher postpartum depressive symptomatology was significantly associated with shorter breastfeeding duration and early exclusive breastfeeding cessation, compared with mothers reporting fewer depressive symptoms (Dias & Figueiredo, 2015; see also Dennis & McQueen, 2009). A narrative synthesis of 33 studies indicated higher postpartum anxiety was associated with reduced likelihood of exclusive breastfeeding and increased risk of early breastfeeding cessation, compared with mothers reporting fewer anxiety symptoms (Fallon, Groves, Halford, Bennett, & Harrold, 2016). High prenatal anxiety was also associated with reduced breastfeeding intention and exclusivity Grigoriadis et al., 2018).
Guilt has also been an associated outcome of infant feeding, and especially so for formula supplementation. Guilt has been defined as feelings of remorse concerning a moral transgression (Niedenthal, Tangney, & Gavanski, 1994). In existing literature, formula feeding was perceived as a moral failing, as maternal discourse was frequently spoken of synonymously with having not done 'right' by one's infant (Brodribb, Fallon, Jackson, & Hegney, 2010;Lakshman & Ong, 2009) and with having failed to meet expectations of oneself postnatally (Kair, Flaherman, Newby, & Colaizy, 2015). Such feelings of guilt have been reportedly exacerbated by breastfeeding education and promotion which inefficiently prepares women for postnatal infant feeding difficulties (Groleau, Pizarro, Molino, Gray-Donald, & Semenic, 2016).
Guilt has also been associated with feelings of anger being held towards healthcare professionals, when mothers perceived that they had received ineffective support (Humphries & McDonald, 2012).
Perceiving that healthcare professionals were promoting breastfeeding as a moral obligation and perceiving that breastfeeding was overly medicalised were both linked with guilt and undermined maternal autonomy (Benoit, Goldberg, & Campbell-Yeo, 2016). Indeed, perceiving that formula feeding was risky to infant health and perceiving that one had moral responsibility over infant feeding method were both associated with feelings of guilt for women who were supplementing with formula (Taylor & Wallace, 2017;Williams, Donaghue, & Kurz, 2012). Interestingly, women who perceived that supplementing with formula milk was not their decision did not experience guilt to the same degree, highlighting the importance of perceived responsibility in determining the presence or absence of maternal guilt (Holcomb, 2017).
Shame also occurs in association with infant feeding experiences.
Shame has been defined as the internalisation of guilt to the self, especially if one perceives themselves to be failing in front of others (Niedenthal, Tangney, & Gavanski, 1994). Although both guilt and shame concern a perceived or actual moral transgression, guilt is externalised and behaviour-orientated, whereas shame concerns the internalisation of said transgression to the self (Niedenthal, Tangney, & Gavanski, 1994). Taylor and Wallace (2012) further supported this definition in finding that globalised assessments of the self as a bad mother, in association with formula feeding practice or public breastfeeding, exceeded the behaviour-focused feelings of guilt and instead focused on the self as a failing entity. In infant feeding literature, feeling that one was failing their moral obligation to breastfeed when challenges were experienced, and feeling like one was failing in front of others, were both linked with feelings of shame (Hanell, 2017).
In quantitative infant feeding literature, guilt has been examined through binary response options 'yes/no' in response to direct questions about feeling guilty due to one's infant feeding method (Chezem, Montgomery, & Fortman, 1997;. To the author's knowledge, there are currently no quantitative studies examining shame in relation to infant feeding outcomes. In qualitative infant feeding literature, guilt and shame have been identified in thematic analysis (e.g., identified theme 'relief and guilt' in Fahlquist, 2016, and'shame' examination in Hanell, 2017) and have occasionally been grouped in thematic analysis (e.g., identified theme, 'stress, shame and guilt' in Asiodu, Waters, Dailey, & Lyndon, 2017).
Framework analyses have also been used to offer a holistic picture of how shame is experienced in an infant feeding context, which have considered both individual vulnerabilities, for example, idealised expectations of 'good mothering', and social factors, for example, fears concerning breastfeeding in public (Thomson, Ebisch-Burton, & Flacking, 2015).
Current literature evidences the relationship between poorer breastfeeding outcomes and negative maternal affect, such as anxiety, depression, guilt and shame, in developed countries. Although there are existing reviews examining the relationship between infant feeding outcomes and maternal anxiety and depression, guilt and shame literature has yet to be synthesised in relation to infant feeding outcomes. Understanding this relationship may allow better identification of women vulnerable to experiencing these emotions and provide recommendations for tailored care. Given the identified decline in breastfeeding prevalence at 6 months postpartum compared with initiation rates in developed countries (AGDH, 2019;CDCP, 2019;Chalmers et al., 2009;McAndrew et al., 2012;Theurich et al., 2019), the current review will synthesise data from developed countries, only. This mixed-methods systematic review aims to (a) examine the relationship between guilt and/or shame and different infant feeding outcomes and (b) examine how guilt and/or shame are experienced differentially depending on infant feeding method.

| METHOD
The current review was preregistered on PROSPERO in November 2018 (https://www.crd.york.ac.uk/PROSPERO/#recordDetails). A protocol was developed based on a scoping literature search.

| Ethical considerations
Ethical approval was not required for the current study as it used secondary data collection and analysis. Findings from this study will form part of LJ's PhD thesis.

| Eligibility criteria
Studies were included if they explicitly explored guilt and/or shame as variables or if they reported them as key themes in an infant feeding context and if they were conducted in developed countries, as defined by the Statistical Annex (Country Classification, 2014).
Given cultural variation in breastfeeding practices and maternal wellbeing between developed (Leahy-Warren, Creedon, O'Mahony, & Mulcahy, 2017) and developing (Wanjohi et al., 2017) countries, it was deemed appropriate to only include studies from the former. This is also supported by the identified decline in breastfeeding prevalence at 6 months postpartum compared with initiation rates reported in developed countries (AGDH, 2019;CDCP, 2019;Chalmers et al., 2009;McAndrew et al., 2012;Theurich et al., 2019). See Table 1 for inclusion criteria for study selection.

| Search strategy
A search strategy was developed in line with Population Exposure Outcomes criteria (PEO; University of London, 2020; see Table 2).
PEO criteria were utilised to develop clear study aims and research questions, as recommended by O'Harhay and Donaldson (2020) and in line with other attempts to answer health-related questions (Davies, 2011). PEO criteria were also utilised to map inclusion criteria for article selection at title, abstract and full text screening stages. Key terms utilised in the final search strategy were determined via a scoping literature search and the subsequent identification of relevant key words included in identified papers. All named authors agreed upon the final search strategy.
Keywords used to search for articles included 'shame*'; 'guilt*'; 'stigma*'; 'moral*'; 'breastfeed*'; 'breast feed*'; 'breast-feed*'; 'bottle feed*'; 'bottle-feed*'; 'infant feed*'; 'infant-feed*'; 'formula feed*'; 'formula-feed*'; 'combi* feed*' and 'human lactat*'. Boolean operators were used to blend keywords, and truncation was used to identify variations of keywords. Articles were screened for suitability against eligibility criteria, outlined in Table 1 The search strategy identified 7 papers which were written in French, 1 study which was written in Polish, 1 study which was written in Spanish and 1 study which was written in Portuguese. Studies not written in English were translated by independent researchers and screened using the outlined search strategy and inclusion criteria. The search strategy was rerun in February 2020 identifying 1 additional article.
During screening, 1 paper was identified which examined a sample of mothers who experienced breastfeeding aversion. It was decided to remove this paper due to associated feelings of shame which may have otherwise confounded findings (Morns, Steel, Burns, & McIntyre, 2020). An additional 2 papers involved samples of women who had a history of sexual abuse. These papers were excluded due to evidence suggesting that historic sexual abuse may affect parenting style and anxieties and may contraindicate breastfeeding comfort due to feelings of shame (Haiyasoso, 2019;Wood & Esterik, 2010). A further paper involved mother-infant dyads who had been separated shortly after birth due to medical emergency. This paper was excluded due to subsequent interruption of breastfeeding initiation in the first hour of giving birth (Phillips, 2013). Finally, 1 study involved a sample of refugee women. This study was excluded due to evidence suggesting that this particularly vulnerable group have exceptionally inadequate access to social and healthcare professional support which may have otherwise confounded findings (Lerseth, 2013;Madanat, Farrell, Merrill, & Cox, 2007). See Figure 1

| Quality assessment
The Standard Quality Assessment Criteria for Evaluating Primary Research Papers from a Variety of Fields (SQAC; Kmet, Cook, & Lee, 2004) was used for quality assessment. The SQAC contains separate point-based checklists for quantitative and qualitative methodologies. Mixed methods studies were assessed using both checklists.
Quality assessment was conducted by 2 researchers independently.
Any discrepancies were discussed, and if agreement could not be reached, then a third member of the research team was consulted.
Quality assessment framed suggestions made for future research.

| Data extraction
Data extraction from the 20 included studies comprised references, aims and/or hypotheses, inclusion and exclusion criteria, sampling method and characteristics, drop-out rate, design, location, methodology, outcome variables, descriptive statistics, analysis method, summary of guilt and/or shame findings, outline of guilt and/or shame definition, secondary findings, related keywords and methodological comments. The following information was extracted from quantitative studies only: control for confounders and exposure/outcome variable(s). Data extraction was conducted by 2 researchers independently. Any discrepancies were discussed, and if agreement could not be reached, then a third member of the research team was consulted.

| Analysis
A narrative synthesis (Rodgers et al., 2009)  T A B L E 2 Population exposure outcomes (PEO) for exploring guilt and shame in relation to infant feeding outcomes Review question(s) Population Exposure Outcome (a) Examine the relationship between guilt, shame, and infant feeding outcomes. (b) Explore how guilt and shame are experienced by mother's, dependent on infant feeding method.
Women who have given birth in the past 6 months to a full-term (>37 weeks), healthy infant (>2,500 g). Absence of maternal or infant congenital abnormalities which would otherwise affect ability to breastfeed. Women with no clinical diagnosis of mental distress, unless controlled for in study analysis. When reported, absence of traumatic experiences e.g. history of sexual abuse, or significant displacement, which may otherwise affect emotional or practical infant feeding experiences.
To be included in the current analysis, included studies needed to involve participants with infants under 6 months of age, who have previously or are currently experiencing postnatal guilt and/or shame. As such, included articles needed to explicitly examine maternal guilt and/or shame in relation to infant feeding outcomes.
Formula and breastfeeding intention, initiation, duration and method at time of investigation. Qualitative experiences related to outcome measures were also explored.
F I G U R E 1 PRISMA 2009 flow diagram identifying three stage systematic screening process for article inclusion

| RESULTS
After removal of duplicates, the search strategy yielded 467 studies dating 1948-2017, across 34 databases (see Table 3 for tabulation of article frequencies by database, before and after removal of duplicates  (n = 192), 2 studies used single unit participation (n = 2) and 1 study consisted of 2 qualitative online surveys, with 5 and 42 participants.
Total qualitative sample size was 388. Given that only 2 included papers examined shame in relation to infant feeding outcomes, and neither of these included papers examined shame quantitatively, guilt and shame were grouped together in the framework synthesis, and results were split by infant feeding method.
Of the 3 included quantitative papers, 2 studies used a crosssectional, online methodology, and 1 study used a longitudinal, telephone questionnaire. Quantitative sample size ranged from 53 to 679. Total quantitative sample size was 1,333. The search strategy identified 5 mixed method studies, 4 of which used the same dataset (Lee, 2007a(Lee, , 2007b(Lee, , 2007cLee & Furedi, 2005) using a structured questionnaire and semistructured interviews. The fifth study involved quantitative analysis of telephone questionnaires and semistructured interviews. Sample size for mixed methods papers ranged from 12 to 33 for qualitative components (n = 45) and 86 to 504 for quantitative components (n = 590). Total sample size for mixed methods studies was 635. See Table 4 for summary table of included literature.

| Narrative synthesis of quantitative and mixed methods studies
Multivariate analyses were given reporting precedence over bivariate, univariate and descriptive analyses reported within included articles with quantitative components. Of the 8 included quantitative papers, only 2 quantitative studies used multivariate analyses.
T A B L E 3 Frequency table to display articles identified from search strategy per database, before (and after) removal of duplicates

| Breastfeeding
There were no significant differences between guilt scores of women with exclusive breastfeeding intentions and women with combination feeding intentions during pregnancy . However, postnatally, risk of guilt was 6 times higher for combination feeders compared with exclusive breast feeders (Adjusted RRR: 0.17, 95% CI: 0.10, 0.27).

| Formula feeding
Risk of guilt was 7 times lower for formula feeding women who had had exclusive formula feeding intentions during pregnancy (Adjusted RRR: 0.14, 95% CI: 0.08, 0.26) and 2 times lower for women with combination feeding intentions (RRR: 0.48, 95% CI: 0.29, 0.79), compared with women who had had exclusive breastfeeding intentions in pregnancy but whom were exclusively formula feeding postpartum . Risk of guilt was 4 times lower for women who had exclusively formula fed since birth (Adjusted RRR: 0.45, 95% CI: 0.25, 0.79), and 2 times lower for combination feeders since birth (Adjusted RRR: 0.38, 95% CI: 0.21, 0.64) compared with women who initiated exclusive breastfeeding but whom were exclusively formula feeding postpartum .
In bivariate analyses, not meeting breastfeeding intentions was associated with significantly higher guilt compared with women who met antenatal goals when returning to work within 1 year postpartum (p = .004; Chezem, Montgomery, & Fortman, 1997). In descriptive analyses, 33% of exclusively formula feeding women with antenatal breastfeeding intentions felt guilty in relation to their infant feeding method (Lee, 2007a(Lee, , 2007b(Lee, , 2007cLee & Furedi, 2005).
3.5 | Examine how guilt and/or shame are experienced differentially depending on infant feeding method Framework synthesis identified 2 major themes from 6 studies for breastfeeding mothers: 'underprepared and ineffectively supported' and 'morality and perceived judgement'. Due to only 2 included studies examining experiences of combination feeding mothers, findings from combination feeding and exclusively breastfeeding mothers were collapsed into the category breastfeeding mothers.

| Underprepared and ineffectively supported
Mothers perceived that health professionals ineffectively prepared them for postpartum breastfeeding challenges and postnatal experiences were consequentially often at odds with prenatal expectations (Fox, McMullen, & Newburn, 2015). This disparity led to feelings of self-doubt, anxiety (Thomson, Ebisch-Burton, & Flacking, 2015) and undermined breastfeeding self-efficacy, 'In the hospital they kept repeating that it shouldn't be painful, if you are doing it right it shouldn't hurt. And that wasn't particularly helpful, because it was painful for me ' (p. 6, Mother;Fox, McMullen, & Newburn, 2015).
Feeling unprepared for breastfeeding challenges also led to feelings of guilt (Asiodu, Waters, Dailey, & Lyndon, 2017) and shame (Hanell, 2017)  Breastfeeding women also felt shame in response to perceptions of overinvolved care and nonconsensual breast handling by healthcare professionals (Thomson, Ebisch-Burton, & Flacking, 2015). Breastfeeding mothers would have instead preferred to receive more hands-off, practical support and also expressed a preference to have received more individualised infant feeding support (Dalzell, 2007

| Mortality and perceived judgement
In most included qualitative studies, breastfeeding mothers felt morally obliged to adhere to 'breast is best' discourse, which was associated with guilt when breastfeeding difficulties were experienced (Fox, McMullen, & Newburn, 2015). Quantitative analysis also identified that guilt was equally likely to be experienced in association with internal as with external factors, with 37.6% of breastfeeding women experiencing internal (feelings of guilt originating from how one feels about their infant feeding method) and 32.7% of breastfeeding women experiencing external (feelings of guilt originating from how one perceives others to feel about their infant feeding method) guilt.
Guilt was, however, also felt through both internal and external channels for 26.7% of mothers . It was identified commonly in the framework synthesis that trying and failing to breastfeed were more morally acceptable than formula feeding from birth, and alternative feeding methods were often perceived as wrongful (Spencer, Greatrex-White, & Fraser, 2014), 'I couldn't help but feel that I was sort of, I wasn't doing my job properly, if I didn't at least give it my absolute best shot' (p. 6, Mother; Fox, McMullen, & Newburn, 2015).
Formula feeding was equated with inadequate mothering (Dalzell, 2007) and was commonly associated with loss of maternal identity (Hanell, 2017)  Indeed, many women feigned effortless breastfeeding experiences, which were often at odds with their actual private experiences, in fear of being judged as a bad mother by healthcare professionals (Spencer, Greatrex-White, & Fraser, 2014) or by family members (Hanell, 2017). Judgemental comments regarding breastfeeding from family and friends (Fox, McMullen, & Newburn, 2015) led to social sphere withdrawal (Thomson, Ebisch-Burton, & Flacking, 2015).
Quantitative analysis also provided evidence for the relationship between guilt and social support networks, with 58.7% of breastfeeding women experiencing external guilt in relation to family and 31.7% experiencing external guilt in relation to other mothers   (p. 241, Veronica;Hanell, 2017).
Breastfeeding mothers resisted seeking help and often spoke of fearing being perceived as a failure. This was often discussed by women experiencing guilt in the context of breastfeeding pressure (Spencer, Greatrex-White, & Fraser, 2014

| Formula feeding mothers
Framework synthesis identified 2 major themes from 9 studies which examined the experiences of formula feeding women: 'frustration with infant feeding care' and 'failures, fears and forbidden practice'.

| Frustration with infant feeding care
Inconsistent guidance and support (Lamontagne, Hamelin, & St-Pierre, 2008) were perceived as frustrating and confusing (Lagan, Symon, Dalzell, & Whitford, 2014), and there was an expressed need for better quality in infant feeding care. Healthcare professionals were quick to blame mothers for breastfeeding difficulties, which led to feelings of guilt for women who were unable to breastfeed and who, subsequently, were formula feeding at the time of investigation (Fahlquist, 2016). Quantitative analysis also found that 64% of formula feeding women experienced external guilt in relation to healthcare professionals . Feeling undermined by healthcare professionals and publicly embarrassed was also mentioned by mothers experiencing guilt, 'I felt awful, my daughter was crying, she didn't eat enough, lost weight, I panicked all the time and didn't know what to do. The child health center told me the problem was mine, I did something wrong … no one helped me, and everyone was just nagging about how good it is to breastfeed' (p. 235, Mother;Fahlquist, 2016).
Lack of respect from healthcare professionals regarding maternal wishes to supplement with formula exacerbated feelings of guilt and shame (Hvatum & Glavin, 2017) and resulted in resentment being held towards healthcare professionals (Murphy, 2000), 'My baby didn't gain in weight but lost 750g, but even then I wasn't allowed to give substitute. I got the understanding that there had to be a complete crisis first. Almost like they had to legalize it. It makes you feel even more unsuccessful ' (p. 3149, Mother 8;Hvatum & Glavin, 2017).
Frustration with quality of care resulted in concealment of infant feeding method and provoked feelings of guilt (Lee, 2007b), 'I was lying a lot, especially with the health visitor because every week … "still breastfeeding?" It got to a stage when I was like, "yeah still, still doing a bit but giving [baby] the formula at night-time." Because it was just the same question and they make you feel guilty' (p. 304, Mother; Lee, 2007b).
3.6.2 | Failures, fears, and forbidden practice Women experiencing guilt often internalised feelings that they were letting their baby down and feared potential infant health consequences of formula supplementation (Fahlquist, 2016), whereas shame was attributed to the self and experienced for seemingly having failed in front of other mothers (Crossley, 2009). Formula feeding often led to dissociation from one's maternal identity (Murphy, 2000) and defensiveness over infant feeding method (Lee & Furedi, 2005).
Failing to breastfeed was also associated with self-blame (Mozingo, Davis, Droppleman, & Merideth, 2000) and postnatal depression (Thomson, Ebisch-Burton, & Flacking, 2015). Quantitative analysis also found that for formula feeding mothers, guilt was experienced more commonly in relation to internal feelings (30%) than in relation to external factors (12%). Guilt was, however, also felt through both internal and external channels for 55% of formula feeding mothers . The following participant accounts reflect these findings that formula feeding was linked with internalised perceptions of the self as having failed to achieve good mothering status, 'It was all "Well, I breast fed for two years," "Well I breastfed for Formula feeding mothers often avoided help-seeking behaviour and frequently spoke of fearing judgement for their infant feeding method from healthcare professionals and social support networks (Crossley, 2009;Lee & Furedi, 2005). Quantitative findings also demonstrated that 68% of formula feeding mothers experienced external guilt associated with other mothers . Prohibition of formula discussions also led mothers to feel that formula feeding was forbidden and that there was pressure to breastfeed (Crossley, 2009;Lee, 2007b), 'The antenatal class I had attended was heavily biased towards breastfeeding. For instance, in the session on feeding, a flip chart was put up and we were asked to list the advantages and disadvantages of feeding babies in particular ways. The midwife only wrote down the advantages of breastfeeding and ignored anyone who mentioned bottle-feeding advantages' (p. 81, in text; Crossley, 2009), 'When no one talks about formula, and the paediatric nurse says that she cannot "promote" formula, you feel like a criminal, like you are doing something illegal' (p. 236, Mother; Fahlquist, 2016).

| DISCUSSION
This mixed methods systematic review aimed to address 2 research questions, 'examine the relationship between guilt and/or shame and different infant feeding outcomes' and 'examine how guilt and/or shame are experienced differentially depending on infant feeding method'. A framework synthesis of qualitative and quantitative data and a narrative synthesis of quantitative data were utilised to address the research questions. The framework synthesis identified 4 key themes: 'underprepared and ineffectively supported', 'morality and perceived judgement' (breastfeeding), 'frustration with infant feeding care' and 'failures, fears and forbidden practice' (formula feeding).
Previous reviews have found depression (Dennis & McQueen, 2009;Dias & Figueiredo, 2015) and anxiety Grigoriadis et al., 2018) to be related to formula supplementation and early breastfeeding cessation. The current review extends this work to other domains of negative effect known to be associated with poorer breastfeeding outcomes, namely, guilt and shame. From a biological standpoint, depression and anxiety (Stuebe, Grewen, & Meltzer-Brody, 2013) are suggested to adversely affect hormones necessary for breastfeeding (Lonstein, 2007). Oestrogen plays an important role in the process of milk ejection during breastfeeding (Uvnäs-Moberg & Eriksson, 1996) and is lowered in women with postnatal depression (Harris, 1996). Similarly, women who do not breastfeed demonstrate elevated cortisol levels, heart rate and lowered oxytocin in response to external stressors, compared with breastfeeding women (Cox et al., 2015). Given the link between shame and postnatal depression in the current review, biological theories underlying the relationship between negative maternal affect and poorer breastfeeding outcomes might extend to include the roles of guilt and shame.  (Fallon, Harrold, & Chisholm, 2019). Although midwives desire to be 'skilled companions', they often find it difficult to provide this support due to resource constraints and work environment barriers (Burns, Fenwick, Sheehan, & Schmied, 2013;Dykes, 2005;Mclelland et al., 2015).

| Morality and perceived judgement
This theme is supported by existing literature highlighting that mothers frequently experience social and societal pressures to breastfeed through synonymous associations with 'good mothering' (Hunt & Thomson, 2017).This can lead to feelings of guilt, failure, fears of being judged and inhibition of help seeking behaviour (Regan & Brown, 2019;Taylor & Wallace, 2017;Williams, Donaghue, & Kurz, 2012;Williams, Kurz, Summers, & Crabb, 2013).
It is therefore important to move away from moral-based language to minimise negative emotions for those experiencing breastfeeding difficulties or early breastfeeding cessation. No quantitative literature examined shame in relation to infant feeding outcomes. This is concerning, given both its associations with negative breastfeeding experiences in qualitative literature, and its associations with postnatal depression and help-seeking avoidance (Dunford & Granger, 2017). Future research should therefore aim to quantify the relationship between maternal shame and infant feeding outcomes.
Although it is important to promote and support breastfeeding, it is also necessary to ensure that formula feeding mothers have adequate emotional and practical support to feed their baby safely and responsively.

| Failures, fears and forbidden practice
Formula feeding mothers who experienced guilt were more prone to feelings of failure which were discussed in the context of 'breast is best' discourse. This may be explained by selfdiscrepancy theory, which proposes that maternal guilt and shame result from discrepancies between one's actual and ideal self (Liss, Schiffrin, & Rizzo, 2012). This suggests a need for a more flexible promotional message which dissipates an 'all or nothing' breastfeeding mentality and instead focuses on a more incremental 'every feed counts' approach to providing breastfeeding support (Braimoh & Davies, 2014;Brown, 2016;Símonardóttir & Gíslason, 2018).

| Limitations
The quality of included studies limited the ability to form firm conclusions. The majority of included quantitative literature did not report statistical analyses in full (Chezem, Montgomery, & Fortman, 1997;Lee, 2007aLee, , 2007bLee, , 2007cLee & Furedi, 2005), and one study lacked scale validity testing , collectively suggesting caution should be taken regarding validity of findings.
Some quantitative papers involved binary examination of guilt .
Binary examination of concepts is problematic as it provides a reductionist view of how guilt and shame are experienced within an infant feeding context. Future research should therefore aim to explore contributing factors and outcomes of guilt and/or shame, to gain a clearer narrative for these negative affective states within an infant feeding context.

| CONCLUSION
A mixed-methods systematic review synthesising the findings from 20 papers examined how guilt and/or shame were related to different infant feeding outcomes and examined how guilt and/or shame were experienced differentially depending on infant feeding method. Quantitative findings suggest guilt is experienced more frequently as breastfeeding exclusivity declines, especially when breastfeeding intentions are unmet. For breastfeeding mothers, guilt was experienced in relation to family and peers, whereas for formula feeding mothers, guilt was experienced in relation to healthcare professionals and peers. Lack of quantitative exploration of shame in relation to infant feeding outcomes prompted suggestions for future research. Qualitative findings identified a need for more realistic, nonjudgemental and mother-centred support to minimise guilt and shame experiences for those who breastfeed. For formula feeding mothers, providing practical support about how to feed safely and providing emotional support to those who are unable to meet their breastfeeding intentions is critical for maternal wellbeing. A shift is also recommended from a '6 months exclusive breastfeeding' to an 'every feed counts' approach to providing breastfeeding support.

ACKNOWLEDGMENTS
Miss Harriet Makin (HM) independently screened 25% of included articles during study selection, to assess interrater reliability. The University of Liverpool funded the lead author's PhD studentship. Data sharing was not applicable to this article, as no new data were created or analysed in this study.

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.

CONTRIBUTIONS
LJ was responsible for study conceptualisation, methodology, data analysis and initial manuscript draft. JH independently quality assessed included articles to examine interrater reliability with LJ. VF independently data extracted included articles to assess interrater reliability with LJ. VF reviewed identified themes based on feasibility. LJ, VF, LDP and JH reviewed, edited and approved the final manuscript as submitted.